Multiple object detection

[Alpkant]

I am using runner to create detection outputs. Even though there are multiple objects of the same class, the model output is always single. Even if I use num_decodes > 1, it gives the same prediction multiple times. Is this a bug from the code side, or is the model not trained to predict multiple objects in the same image?